Laluet et al. (2024) Drainage assessment of irrigation districts: on the precision and accuracy of four parsimonious models

Identification

• Journal: Hydrology and earth system sciences
• Year: 2024
• Authors: Pierre Laluet, Luis Olivera-Guerra, Víctor Altés, Vincent Rivalland, Alexis Jeantet, Julien Tournebize, Omar Cenobio-Cruz, Anaïs Barella-Ortiz, Pere Quintana Seguí, Josep María Villar Mir, Olivier Merlin
• DOI: 10.5194/hess-28-3695-2024

Research Groups

• Centre d'Etudes Spatiales de la Biosphère (CESBIO), Université de Toulouse, CNES, CNRS, IRD, UPS, Toulouse, France
• Department of Geodesy and Geoinformation, TU Wien, Vienna, Austria
• Laboratoire des Sciences du Climat et de l'Environnement, CEA-CNRS-UVSQ-UPSACLAY, UMR 8212, IPSL, Gif-sur-Yvette, France (now at)
• isardSAT, Barcelona, Spain
• Department of Environment and Soil Sciences, University of Lleida, Lleida, Spain
• UR HYCAR, University of Paris-Saclay, INRAE Jouy-en-Josas, Antony, France
• Observatori de l'Ebre, Universitat Ramon Llull – CSIC, Roquetes, Spain

Short Summary

This study assesses the precision (site-calibrated performance) and accuracy (default parameter performance) of four parsimonious drainage models combining two surface (RU, SAMIR) and two subsurface (Reservoir, SIDRA) components in a semi-arid irrigated district. The RU-Reservoir model demonstrated the highest precision (average KGE($Q^{0.5}$) of 0.87) when calibrated, while the SAMIR-Reservoir model provided the most consistent rough estimates of drainage dynamics and amounts using default parameters.

Objective

• Can parsimonious models with different degrees of complexity precisely reproduce the daily drainage in a semi-arid irrigated context with site-specific calibration?
• Can such models with default calibration (with parameter values provided in the literature) reproduce drainage quantities and dynamics, even roughly?

Study Configuration

• Spatial Scale: Two sub-basins (AB1: 116 hectares, AB2: 2050 hectares) within the Algerri–Balaguer irrigation district, northeastern Spain.
• Temporal Scale: Daily time step, covering 21 months (February 2021 to October 2022) for AB1 and 18 months (May 2021 to October 2022) for AB2.

Methodology and Data

• Models used: Four combined parsimonious models:
  • RU-Reservoir (two main parameters)
  • RU-SIDRA (three main parameters)
  • SAMIR-Reservoir (three main parameters)
  • SAMIR-SIDRA (four main parameters) (RU: simplified FAO-56 water balance; SAMIR: FAO-56 dual crop coefficient model; SIDRA: semi-analytical Boussinesq equation model; Reservoir: conceptual linear depletion model).
• Data sources:
  • Observation: Continuous daily drainage discharge measurements (CTD-10 sensors) at the outlets of AB1 and AB2.
  • Irrigation: Daily flow data from the pumping station (SAIH Ebro Basin), aggregated weekly.
  • Meteorological: Spatial average of precipitation and reference evapotranspiration ($ET_0$) from five stations (Catalan Meteorological Station Network).
  • Soil: SoilGrids product (250 m resolution) for soil texture (silty clay loam).
  • Remote Sensing: Sentinel-2 derived Normalized Difference Vegetation Index (NDVI) (10 m resolution, 5 day revisit time) used by SAMIR.
  • Calibration: Multi-objective Non-dominated Sorting Genetic Algorithm (NSGA-II) using Kling–Gupta efficiency ($KGE(Q^{0.5})$) for drainage and Root-Mean-Square Deviation (RMSD) for irrigation.

Main Results

• Precision Evaluation (Site-Calibrated):
  • RU-Reservoir was the most precise model, achieving an average $KGE(Q^{0.5})$ of $0.87$.
  • SAMIR-Reservoir followed with an average $KGE(Q^{0.5})$ of $0.79$.
  • The Reservoir-based models outperformed SIDRA-based models due to the low responsiveness (smooth, low peaks) of the studied hydrosystems, which better matched the Reservoir model's dynamics.
  • Calibrated parameter values showed significant variability between the 2021 and 2022 periods, indicating limited predictive capacity (robustness) of the semi-empirical models.
• Accuracy Evaluation (Default Parameters):
  • All 16 model-sub-basin-period combinations showed unsatisfactory $KGE(Q^{0.5})$ values (below 0.5).
  • SAMIR-Reservoir provided the most consistent rough estimates of drainage dynamics and amounts compared to the other three models when using default parameters from the literature.
  • RU-based models failed to simulate discharge in 2021 with default parameters because their lack of spatialization and reliance on average irrigation required lower-than-literature $S_{inter}$ values to generate sufficient recharge.

Contributions

• First assessment of the precision and accuracy of four parsimonious drainage models (RU-Reservoir, RU-SIDRA, SAMIR-Reservoir, SAMIR-SIDRA) in a semi-arid irrigated environment.
• Demonstrated that the simplest model (RU-Reservoir, two parameters) achieved the highest precision when calibrated, suggesting that increased descriptive complexity (e.g., SAMIR's detailed ET simulation or SIDRA's physical basis) does not necessarily translate to higher precision in this specific low-responsive environment.
• Identified SAMIR-Reservoir as the most robust model for providing rough drainage estimates in ungauged irrigated sub-catchments using only default literature parameters.
• Highlighted the critical need for spatialized models or accurate plot-scale irrigation data when applying these models in heterogeneous irrigated districts.

Funding

• IDEWA project (grant no. ANR-19-P026-003) of the Partnership for Research and Innovation in the Mediterranean Area (PRIMA) program.
• European Horizon 2020 ACCWA project (grant agreement no. 823965) in the context of the Marie Skłodowska-Curie research and innovation staff exchange (RISE) program.

Citation

@article{Laluet2024Drainage,
  author = {Laluet, Pierre and Olivera-Guerra, Luis and Altés, Víctor and Rivalland, Vincent and Jeantet, Alexis and Tournebize, Julien and Cenobio-Cruz, Omar and Barella-Ortiz, Anaïs and Quintana‐Seguí, Pere and Mir, Josep María Villar and Merlin, Olivier},
  title = {Drainage assessment of irrigation districts: on the precision and accuracy of four parsimonious models},
  journal = {Hydrology and earth system sciences},
  year = {2024},
  doi = {10.5194/hess-28-3695-2024},
  url = {https://doi.org/10.5194/hess-28-3695-2024}
}

Generated by BiblioAssistant using gemini-flash-latest (Google API)